It is well known that silicon in its various structural configurations (single crystal, multicrystalline, amorphous, micro-nanocrystalline) supplies almost 90% of the substrates used in the photovoltaic industry. It is also known, since years, that the photovoltaic (PV) industry shows a marked growth trend, which demanded and demands a continuous, huge increase of the bulk silicon supply in the order of 30%/yr. In order to fulfill their today- and future needs, many companies worldwide took the decision to start the installation of many thousand tons/year plants, most of them using the Siemens process, some of them using the MG route, to produce the so called solar grade (SG) silicon. The advantages of the Siemens process are well known, as it provides ultrapure silicon, directly usable for growing either single crystalline Czochralski ingots or multicrystalline ingots using the directional solidification (DS) technique. The disadvantages are its high energetic cost (a minimum of 120 kWH/kg) and the possible losses of chlorinated gases in the atmosphere, with possible severe environmental problems. The advantages of the MG route are still potential, as there is no commercially available production of solar silicon as yet, and rely on its reduced energetic costs (a maximum of 25-30 kWh/kg) for a feedstock directly usable for growing multicrystalline ingots using the DS technique. The drawbacks of silicon of MG origin are its larger concentration of metallic impurities, as compared with the Siemens one, the higher B and P content, and the potentially high carbon content. The aim of this paper is to deal with some of the problems encountered so far with the silicon of MG origin with respect to the metallic and non-metallic impurities content, as well as to propose technologically feasible solar grade feedstock specifications. (C) 2010 Elsevier B.V. All rights reserved.